This invention relates to image sensors, and, for example, relates to an image sensor that corrects imaging signals read out from a grid of pixel circuits arranged in columns and rows using a correction value generated on a row-by-row basis.
Image sensors generate an imaging signal having a voltage level proportional to the exposure on a photoelectric conversion element provided in a pixel circuit, and generate pixel data by subjecting an analog-digital conversion process or other processes to the imaging signal. This conversion from the imaging signal to the pixel data is made through an analog-digital conversion processing path independently dedicated to each row.
However, circuits making up the analog-digital conversion processing path have characteristics variations, such as input offset variations, and gain variations. The circuit characteristics variations result in fixed pattern noise that causes color shift of pixels in specific rows. An example method for eliminating the fixed pattern noise is disclosed in Japanese Unexamined Patent Application Publication No. 2008-252605.
The solid-state imaging device disclosed in Japanese Unexamined Patent Application Publication No. 2008-252605 uses a first pixel row in an optical black pixel area, which is located on an upper side of a pixel array, as a first correction pixel row for obtaining a correction value of gain×1, and a second pixel row as a second correction pixel row for obtaining a correction value of gain×8. The solid-state imaging device then applies pixel signals obtained from respective pixels on the first and second correction pixel rows to a column processing unit, thereby obtaining correction values of gain×1 and gain×8 for correcting gain errors between the P phase and the D phase, as well as performing correction processing for the gain errors using these correction values with a multiplexer.